β-lactoglobulin B (βLG) from bovine milk (18.2 kDa) (#L8005) and bovine serum albumin (BSA) (66.5 kDa) (#A7906) were purchased from MilliporeSigma, and used as negative control proteins. Rabbit anti-amyloid-β (#8243), mouse-anti-amyloid precursor protein (APP) (#2452), goat anti-rabbit IgG-HRP-linked (#7074) and horse anti-mouse IgG-HRP-linked (#7076) antibodies were obtained from Cell Signaling Technology. Affinity purified anti-FAIM antibody was obtained from rabbits immunized with CYIKAVSSRKRKEGIIHTLI peptide (located near the C-terminal region of FAIM) as previously described (Kaku and Rothstein, 2009a,b).

The Neuro 2A (N2a) cell line was obtained from the American Type Culture Collection (ATCC, #CCL-131). N2a cells were cultured in DMEM/F12 medium (Cytiva) containing 10% FCS, 15 mM HEPES, 2.5 mM L-glutamine and 0.1 mg/ml penicillin and streptomycin. Transfection was performed using Lipofectamine 2000 according to the manufacturer’s instructions (Invitrogen). In brief, N2a cells were seeded at 1 × 10⁵ cells/well in 24-well plates, and transfected 24 h later with the indicated DNA plasmid (0.5 μg/well) using Lipofectamine 2000 transfection reagent (1 μl/well). pcDNA FRT TO- APP695 and pcDNA FRT TO- APPSwed/Ind were gifts from Aleksandra Radenovic (Addgene plasmids #114193 and #114194, respectively). pcDNA3.3 plasmid (Themo Fisher Science, #K830001) was obtained from Invitrogen, and was used as a negative control (Empty).

Guide RNA (gRNA) sequences for mouse faim gene were designed using a CRISPR target design tool¹ in order to target the exon after the start codon as previously described (Kaku and Rothstein, 2020). Briefly, designed DNA oligo nucleotides were ligated into pSpCas9(BB)-2A-GFP (PX458) vector (Addgene plasmid #48138, a gift from Feng Zhang) at the Bpi1 (Bbs1) restriction enzyme sites using the “Golden Gate” cloning strategy. The presence of insert was verified by sequencing. Empty vector was used as a negative control. After fourth passage, transfection was performed using lipofectamine 2000 as above and a week after the transfection, eGFP⁺ cells were single-sorted with a Melody instrument (Becton Dickinson), and seeded into 96 well plates. FAIM knockout clones were screened by limiting dilution and western blotting. The cells were expanded for three passages and afterward frozen in 90% FBS and 10% DMSO (Sigma-Aldrich, # 472301) at passage seven in liquid nitrogen for future use. After thawing cells, the cells with total 10 passages were used for the further experiments.

Neuro 2A cells were detached by Trypsin-EDTA. Adherent and floating cells were harvested and pooled, after which cells were resuspended in 2 μg/ml 7-aminoactinomycin D (7-AAD) (Anaspec, #AS-83201). Cell viability was assessed using an Attune flow cytometer (Thermo Scientific). Data were analyzed using v10 software (TreeStar).

His-tag protein expression vectors were constructed using pTrcHis TA vector (Themo Fisher Science, #K441001) according to the manufacturer’s instructions as previously described (Kaku et al., 2020). In brief, PCR amplified target genes were TA-cloned into the pTrcHis TA vector (Themo Fisher Science) and inserted DNA was verified by sequencing (Genewiz). Oligonucleotides used in this work for PCR-cloning into pTrcHis TA vector were as follows: human FAIM-S forward, “ATGACAGATCTCGTAGCTGT TTGG”; human FAIM-S reverse, “TTAACTTGCAATCTCTG GGATTTC”; human FAIM-L forward, “ATGGCATCTGG AGATGACAGTC”; human FAIM-L reverse, “TTAACTTGCAA TCTCTGGGATTTC”; human HSP27 forward, “ATGACC GAGCGCCGCGTCCCCTT”; human HSP27 reverse, ‘’TTA CTTGGCGGCAGTCTCATCGGAT.”

The sequence primer used for the pTrcHis-TA vector is as follows: TATGGCTAGCATGACTGGT.” Proteins were expressed in TOP10 competent cells (Themo Fisher Science, #C404010) with IPTG (Invitrogen, #15529019) at 1 mM for 2.5 h at 37°C, and were purified using a Nuvia IMAC Nickel-charged column (Bio-Rad) on an NGC Quest chromatography system (Bio-Rad). Proteins underwent dialysis against PBS using Slide-A-Lyzer Dialysis Cassettes, 10K MWCO (Themo Fisher Science, #66380). Protein purity was verified using TGX Stain-Free gels (Bio-Rad) on ChemiDoc Touch Imaging System (Bio-Rad) and each protein was determined to be >90% pure (Supplementary Figure 1). Random mutagenesis was performed using GeneMorpho II Random Mutagenesis Kit (Agilent, #200550) according to the manufacturer’s recommendation. Briefly, pTrcHis TA vector containing human FAIM-S gene was used as a template, and was amplified for 30 cycles using mutazyme II. Oligonucleotides used in PCR-mutagenesis were as follows: human FAIM-S forward, “GATGACGATAAGGATCCAACCCTTATG”; human FAIM-S reverse, ‘’AGCTTCGAATTGAATTCGCCCTTTTA.” After PCR-amplified DNA was re-inserted into the pTrcHis TA vector, the insert was verified by sequencing, and proteins were expressed as above.

Fibril/aggregate formation of 5 μM Aβ (1–42) (Aβ42) (Anaspec, #AS-72216), assembled in a 384-well clear bottom plate (Corning), was assessed by 50 μM Thioflavin T (ThT) (MilliporeSigma, #T3516) fluorescence using a Synergy Neo2 Multi-Mode Microplate Reader (Bio-Tek). Reader temperature was set at 37°C with continuous shaking between reads. ThT fluorescence intensity was measured using an excitation wavelength of 440 nm and an emission of 482 nm. PMT gain was set at 80. Fluorescence measurements were made from the bottom of the plate, with the top being sealed with an adhesive plate sealer to prevent evaporation.

Preformed Aβ42 fibrils were assembled by incubating 5 μM Aβ42 monomers in amyloid-β assay buffer (Anaspec) for 1 h with agitation at 559 rpm at 37°C. Fibrils were recovered by centrifugation at 14,000 × g for 10 min, washed and resuspended in PBS. Fibril generation was confirmed by 50 μM ThT fluorescence. Fibrils were diluted to the requisite concentration for subsequent disaggregation reactions.

Aβ42 (0.5 μM) pre-formed fibrils were incubated with indicated recombinant proteins at 37°C for 2.5 h. Then, fibril status was determined either by ThT fluorescence, by filter trap assay (FTA), or by analysis of supernatants and pellets after sedimentation and western blotting (sedimentation analysis). For FTA, samples underwent vacuum filtration through a pre-wet 0.2 μm pore size nitrocellulose membrane (GE Healthcare) using a 96 well format Dot-Blot apparatus (Bio-Rad). The membrane was washed twice with PBS and western blotting using anti-Aβ antibody was carried out to detect Aβ fibrils.

Total protein samples were subjected to sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) on an Any kD Mini-PROTEAN TGX Precast Protein Gel (Bio-Rad) followed by immunoblotting after wet transfer for 1 h to PVDF membrane (Bio-Rad) and blocking with 1% non-fat dry milk (Bio-Rad) in TBS-T. In some experiments (Figures 3D, 4A), protein samples were centrifuged at 14,000 × g for 10 min. Pellets were washed by PBS, and suspended in 1x Laemmli buffer. Both supernatant and pellet fractions were subjected to SDS-PAGE. Densitometry quantification of the bands was analyzed by ImageLab software (version 6.1.0) (Bio-Rad). For Supplementary Figure 1, 2, stain-free technology was used to detect total proteins using the ChemiDoc Touch Imaging System (Gilda and Gomes, 2013).

All quantitative data are expressed as mean ± SEM. A two-tailed, unpaired Student’s t-test was used for statistical determinations with GraphPad Prism 8 and 9 software. Values of p < 0.05 are considered statistically significant (*p < 0.05, ^**p < 0.01 or ^***p < 0.001.

We previously showed ubiquitinated protein aggregates induced by cellular stress and mutant SOD1 aggregates accumulate in FAIM-deficient cells (Kaku et al., 2020; Kaku and Rothstein, 2020). To assess the role FAIM plays in prevention of Aβ aggregation, we employed an APP-overexpression system using Neuro 2A cells, a mouse neuroblastoma. It is reported that mutant APP such as Swedish mutant (K670N/M671L) and Indiana mutant (V717F) produce extracellular Aβ fibrils/aggregates due to abnormal processing by endogenous secretase pathways in Neuro 2A cells (Ghosh and Osswald, 2014). We transiently overexpressed APPswe/ind and control APP695 into WT and FAIM-deficient Neuro 2A cells (Figure 1A), and found that cell cultures with mutant APP contained aggregated Aβ 4 days after the transfection (Figure 1B) as previously reported (Park et al., 2015; Cuddy et al., 2017). Regardless of transfection efficiency of APP (Figure 1A), the levels of Aβ fibrils/aggregates in the culture supernatant were dramatically higher in FAIM-deficient Neuro 2A cells than in WT Neuro 2A cells (Figure 1B). The loading control for Figure 1B is shown in Supplementary Figure 2. Although we unexpectedly observed monomer and oligomer Aβ bands in the pellet-insoluble fraction in addition to mature fibrils, mature fibrils might be broken down to monomers/oligomers during the procedure (i.e., exposure to SDS) (Figure 1B). To exclude the possibility that the differences observed in Figure 1B could be due to differences of cell viability after transfection of APP, cell viability was assayed by 7-AAD staining. We did not observe a significant difference in cell death among groups (Supplementary Figure 3). These results demonstrate the essential role of FAIM in altering the fate of mutant APP-derived Aβ.

To examine whether FAIM directly inhibits the generation of Aβ42 fibrils and oligomers in an in vitro cell-free system, and whether the FAIM protein alone is sufficient to inhibit protein aggregation without additional cellular components, we mixed FAIM and Aβ42 monomers and incubated the mixture at room temperature. It has been reported that a mixture of soluble Aβ42 protofibrils and oligomers are formed when monomers are incubated at room temperature without agitation (Jiang et al., 2019). We used HSP27 as a positive control, because HSP27 has been shown to inhibit Aβ42 fibrillization/aggregation in cell-free systems (Lee et al., 2006; Raman et al., 2005; Fonte et al., 2008). We found that the formation of these soluble species was prevented in the presence of recombinant FAIM or HSP27 in this condition (Figures 2A,B). These data suggest that FAIM has activity in preventing the generation of Aβ42 oligomers and fibrils.

To examine whether FAIM inhibits protein Aβ fibrillization we used Thioflavin T (ThT), a dye that only fluoresces when incorporated into amyloid fibrils (Biancalana and Koide, 2010; Malmos et al., 2017; Aliyan et al., 2019). We mixed recombinant FAIM and aggregation-prone monomeric Aβ42 peptide, and monitored aggregation status in real-time by ThT fluorescence intensity. We found that Aβ42 fibrillization was abrogated in the presence of recombinant FAIM or HSP27, but not by PBS, βLG, or BSA (Figure 3A). The normalized ThT fluorescence at 2hr is shown in Supplementary Figure 4. We performed random mutagenesis to identify a mutant form of FAIM-S lacking the activity, and found that S155N mutation abolished the activity of FAIM-S. The FAIM-S S155N mutant protein was used as a negative control in addition to BSA and βLG (Supplementary Figure 4). To confirm these results, fibrillization status was assessed by SDS-PAGE and western blotting (WB). We found aggregated Aβ42 in the high molecular weight range of negative controls (PBS, βLG, and BSA controls). We found the formation of high molecular weight aggregates was dramatically reduced in the presence of recombinant FAIM or HSP27 (Figures 3B,C). We further assessed the solubility of the samples by sedimentation in order to separate soluble protofibrils from mature insoluble fibrils. Both of them were dramatically reduced in the presence of recombinant FAIM or HSP27 (Figures 3D,E).

In order to examine whether FAIM directly inhibits Aβ fibrillization/aggregation and whether FAIM protein is sufficient to inhibit Aβ fibrillization/aggregation by itself without any additional components, we added purified recombinant human FAIM proteins into the cell culture 1 day after transfection with the expression vector for mutant APPswe/ind. Supernatant was collected and separated into soluble and insoluble-pellet fractions as performed in Figure 1. We found that the addition of recombinant FAIM proteins dramatically reduced oligomerized/fibrillized Aβ levels in the insoluble-pellet fraction (Figures 5A,B). Furthermore, the levels of Aβ fibrils/aggregates in the detergent-soluble and insoluble fractions from the cell lysates were also higher in FAIM-deficient Neuro 2A cells than in WT Neuro 2A cells, and the levels of Aβ fibrils/aggregates were reduced by the addition of FAIM proteins (Supplementary Figure 5). These data suggest that extracellular FAIM proteins can prevent the formation of both extracellular and intracellular Aβ oligomers/fibrils produced by Neuro 2A cells overexpressing disease-associated mutant APP.